CN103218916B - Method and system for detecting red light running based on complex high-dynamic environmental modeling - Google Patents
Method and system for detecting red light running based on complex high-dynamic environmental modeling Download PDFInfo
- Publication number
- CN103218916B CN103218916B CN201310117993.6A CN201310117993A CN103218916B CN 103218916 B CN103218916 B CN 103218916B CN 201310117993 A CN201310117993 A CN 201310117993A CN 103218916 B CN103218916 B CN 103218916B
- Authority
- CN
- China
- Prior art keywords
- background
- red light
- vehicle
- crossing
- modeling
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 29
- 238000004387 environmental modeling Methods 0.000 title 1
- 238000001514 detection method Methods 0.000 claims abstract description 26
- 238000005286 illumination Methods 0.000 claims abstract description 26
- 230000033001 locomotion Effects 0.000 claims description 26
- 230000033228 biological regulation Effects 0.000 claims description 4
- 238000004364 calculation method Methods 0.000 claims description 3
- 230000000052 comparative effect Effects 0.000 claims description 3
- 230000007797 corrosion Effects 0.000 claims description 3
- 238000005260 corrosion Methods 0.000 claims description 3
- 238000004321 preservation Methods 0.000 claims description 3
- 239000000203 mixture Substances 0.000 abstract description 4
- 238000010586 diagram Methods 0.000 description 4
- 238000011161 development Methods 0.000 description 2
- 230000018109 developmental process Effects 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 239000011159 matrix material Substances 0.000 description 2
- 230000008929 regeneration Effects 0.000 description 2
- 238000011069 regeneration method Methods 0.000 description 2
- 230000003044 adaptive effect Effects 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 230000006698 induction Effects 0.000 description 1
- 238000012423 maintenance Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 230000001172 regenerating effect Effects 0.000 description 1
- 230000003068 static effect Effects 0.000 description 1
- 230000002459 sustained effect Effects 0.000 description 1
- 230000002123 temporal effect Effects 0.000 description 1
Landscapes
- Image Analysis (AREA)
- Traffic Control Systems (AREA)
Abstract
本发明公开了一种基于复杂高动态环境建模的闯红灯检测方法及系统，包括以下步骤：1）路口画面进行标定；2）利用高斯混合模型对路口画面进行背景建模；3）对建模的背景进行光照检测与分析，并进行背景的更新，得到当前路口的准确背景图像；4）通过背景差分以及跟踪算法处理得到路口车辆的跟踪信息；5）根据车辆的跟踪信息和标定图像判断车辆是否闯红灯。基于此方法的系统包括摄像装置、控制装置，摄像装置与控制装置连接向其发送拍摄到的路口画面，控制装置用于通过高斯混合模型对拍摄到的路口画面进行背景建模，对建模的背景进行光照检测与分析，判断车辆是否闯红灯。本发明能快速、准确地更新背景，适应各种复杂环境，提高闯红灯检测的准确性和适应性。
The invention discloses a detection method and system for running a red light based on complex and high dynamic environment modeling, comprising the following steps: 1) Calibrating the intersection picture; 2) Using a Gaussian mixture model to perform background modeling on the intersection picture; 3) Modeling Perform illumination detection and analysis on the background, and update the background to obtain the accurate background image of the current intersection; 4) Obtain the tracking information of vehicles at the intersection through background difference and tracking algorithm processing; 5) Judge the vehicle according to the tracking information and calibration image of the vehicle Whether to run a red light. The system based on this method includes a camera device and a control device. The camera device is connected to the control device to send the captured intersection pictures to it. The control device is used to perform background modeling on the captured intersection pictures through the Gaussian mixture model, and model the The background performs light detection and analysis to determine whether the vehicle has run a red light. The invention can quickly and accurately update the background, adapt to various complex environments, and improve the accuracy and adaptability of red light detection.
Description
Technical field
The invention belongs to traffic image process field, espespecially a kind of make a dash across the red light detection method and system based on complicated high dynamic environment modeling.
Background technology
The traffic problems that economic development expedites the emergence of become the total difficult problem in city already.China is the developing country of a sustained economic development, and urbanization and motorize develop very swift and violent.In order to solve the series of problems brought thus, needing the input increasing means of transportation, accelerating the construction of transportation supplies; The more important thing is and want the scientific and reasonable existing road traffic facility of use, play their maximum effects.And traditional employing ground induction coil mode detected of carrying out making a dash across the red light has cost high, the shortcoming of Maintenance Difficulty, recently the video detection mode expedited the emergence of obtains crossing video by industrial camera and carries out detection of making a dash across the red light, in order to obtain the vehicle in shot by camera image, general employing has inter-frame difference and background difference two kinds of modes.
Frame differential method algorithm realization is simple, and program design complexity is low, not too responsive to scene changes such as light, can adapt to various dynamic environment, and stability is better, is also at present in the method that the field of intelligent monitoring for complicated high dynamic environment is used often.But it can not extract the complete area of object, border can only be extracted; Depend on the inter frame temporal interval of selection simultaneously.To the object of rapid movement, if the time interval is bigger than normal, when object does not have overlap in the frame of front and back two, two objects separated can be detected as; And the object to microinching, if the time interval is less than normal, when object is almost completely overlapping in the frame of front and back two, then can't detect object, be unsuitable for the detection of making a dash across the red light at crossing.
Background subtraction divides and comprises static background and the background modeling method based on Gauss's body.A frame can be adopted in former old-fashioned supervisory system there is no the picture foreground detection method as a setting of vehicle.The defect of this method is certain clearly, even if because indoor environment all can run into the disturbed condition of light change.And be comparatively popular based on the background modeling method of Gaussian mixture, each pixel in image is carried out modeling by it, the distributed model defining each pixel is the set be made up of multiple single Gauss model, according to certain criterion, the pixel value Renewal model parameter new according to each, judges which pixel is background dot, which is as foreground point.When illumination occurs to change rapidly on a large scale, mixed Gauss model will be its newly-built Gauss's body, but still with former pixel value as a setting (because " strength " of new Gauss's body is less than the stage that can replace original main Gauss's body), until after certain frame number, new Gauss's body replaces original background.But for dynamic, illumination variation complex environment high in city, just there will be background and also do not have enough time to upgrade the situation that complete environment changes again, just there is the result of constantly building new Gauss's body, having to run around all the time wears him out between various change in mixed Gauss model, the object not reaching in real time, accurately monitor.
Summary of the invention
For prior art Problems existing, an object of the present invention is to provide one can upgrade background quickly and accurately, adapt to various complex environment, improve the accuracy and the adaptive detection method of making a dash across the red light based on complicated high dynamic environment modeling of making a dash across the red light and detecting.Another object of the present invention is to provide a kind of detection system based on above-mentioned detection method of making a dash across the red light.
For achieving the above object, the detection method of making a dash across the red light based on complicated high dynamic environment modeling of the present invention, comprises the following steps:
1) the crossing picture that video camera photographs is demarcated;
2) gauss hybrid models is utilized to carry out background modeling to the crossing picture photographed;
3) illumination examination and analysb is carried out to the background of modeling, and carry out the renewal of background, obtain the accurate background image at current crossing;
4) trace information of crossing vehicle is obtained by background difference and track algorithm process;
5) judge whether vehicle makes a dash across the red light according to the trace information of vehicle and uncalibrated image.
Further, step 2) in gauss hybrid models can obtain accurate background image by the method detecting and eliminate a large amount of disturbed motion object.
Further, the detection method of disturbed motion object is specially: the pixel value difference of present frame and background image same point and motion threshold limit value are compared, according to comparative result determination disturbed motion object, then the pixel being defined as disturbed motion object is carried out all disturbed motion targets that piecemeal process obtains in background image.
Further, the background image being defined as disturbed motion target does not carry out context update.
Further, gauss hybrid models carries out background modeling according to hsv color model, context update speed can adjust according to the similarity of current background model and previous background model, if current background model is close to previous model, the renewal speed of the background that then slows down, otherwise the renewal speed accelerating background.
Further, in step 3), the background of modeling is carried out to the method for illumination examination and analysb,
Comprise the steps: 1) edge of background image is extracted, edge is expanded; 2) non-edge of background image is divided into multiple subregion according to position; 3) get two sample points in each subregion, and calculate the quantity of illumination in each subregion; 4) judge whether to carry out subregional context update according to the result of calculation of the quantity of illumination.
Further, step 4) is specially: binary image current frame image and background image difference being obtained present frame prospect, filtered out the interference of people and bicycle by twice expansion and twice corrosion, then use Meanshift track algorithm vehicle to be followed the tracks of to the track of vehicle obtaining current frame image to prospect.
Red light overriding detection system based on complicated high dynamic environment modeling of the present invention, comprise camera head, control device, camera head is connected with control device and sends to it crossing picture photographed, control device is used for carrying out background modeling by gauss hybrid models to the crossing picture photographed, illumination examination and analysb is carried out to the background of modeling, and carry out the renewal of background, obtain the accurate background image at current crossing, the trace information of crossing vehicle is obtained by background difference and track algorithm process, judge whether vehicle makes a dash across the red light according to the trace information of vehicle and uncalibrated image.
The context update mode of the present invention to the gauss hybrid models of classics is improved, no longer the background of each two field picture is all upgraded, but upgrade selectively according to rule, to effectively prevent at height dynamically, under illumination variation complex environment, because mixed Gauss model modeling speed does not reach illumination variation speed, cause the problem that cannot in real time, accurately monitor.
Accompanying drawing explanation
Fig. 1 is the schematic diagram of red light overriding detection system of the present invention;
Fig. 2 is the process flow diagram of the detection method of making a dash across the red light based on complicated high dynamic environment modeling of the present invention;
Fig. 3 is the process flow diagram in gauss hybrid models modeling, the background of modeling being carried out to illumination examination and analysb.
Embodiment
The invention provides a kind of red light running detection record method based on complicated high dynamic environment modeling and a system based on the method, the hardware of whole system forms schematic diagram as shown in Figure 1, comprises industrial camera (containing light compensating lamp), industrial computer.The industrial computer having carried software is the main body of system, and video camera is connected with industrial computer by netting twine.Camera case is arranged on hack lever, and industrial computer is arranged in uphole equipment case.
As shown in Figure 2, the red light running detection method based on complicated high dynamic environment modeling of the present invention, comprises the following steps:
1) demarcate
First, the picture that will photograph video camera does a demarcation, considers that the position of video camera is fixing, only needs to do once to demarcate just can use always.Need at picture acceptance of the bid outlet region, traffic lights region and three lines for the preservation of violation evidence.
2) gauss hybrid models modeling
For the environment of this more complicated in crossing, often there are a large amount of mobile humans or thing in video.In order to adapt to this environment, obtain better background by the method detecting and eliminate a large amount of moving object.
Use I
x, y, trepresent the pixel that present frame coordinate is the point of (x, y), use I
x, y, t-1represent the pixel of previous frame same point, the amount of exercise so put is defined as
M
x,y,t=|I
x,y,t-I
x,y,t-1|
Define a motion threshold limit value Threshold, by amount of exercise binaryzation:
Wherein, B
x, y, twhether to represent present frame coordinate be the point of (x, y) is prospect, works as B
x, y, twhen=1, represent that this point is foreground moving object, otherwise be not foreground moving object.Wherein motion threshold limit value is defined as
Wherein
matrix I
t1 norm, N is the quantity of picture frame mid point, and K is a constant, as K=2, can obtain good effect.
Like this, the object of motion is just determined.Then, foreground point is divided into many pieces, these blocks just represent the target of all motions.
Namely be become large after prospect into
Wherein, BL
x, y, twhether to represent coordinate in the prospect after expansion be the point of (x, y) is prospect, works as BL
x, y, twhen=1, represent that this point is foreground moving object.Store n the value of closing on, for upgrading background pixel:
UpdateBackground
x,yrepresent this point and whether carry out context update (false represent do not upgrade, true represents renewal), t is the time interval of amount of exercise movable information, and T is the current moment.
Background modeling is carried out according to hsv color model, context update speed can adjust according to the similarity of current background model and previous background model, if current background model is close to previous model, then the renewal speed of the background that slows down, otherwise the renewal speed accelerating background.
Wherein, BackgroundingSpeed
x,yrepresent the renewal speed of the point that coordinate is (x, y), the larger renewal speed of numeral is slower; H
x,y, S
x,y, V
x,yh, S, V color component of to be coordinate the be point of (x, y), H
threshold1, H
threshold2, S
threshold1, S
threshold2, V
threshold1, V
threshold2being the threshold value of manually specifying, is namely that just slow down renewal speed if this of current background model is in the threshold value that this point of former background model fluctuates.
3) regeneration of illumination detection and background
Because edge or angle point can not the Lighting information of representative picture, or even the burr of marginal portion is too many, is not suitable for for asking for most illumination.Therefore picture is divided into edge and non-edge part calculates respectively.
First, need to extract the edge of background image.Then edge carries out " overstriking ", namely directly adopts Expanded Operators, is expanded at edge.Like this, the non-edge part that just can not expand other processes.
Following step be experienced by the process of non-edge part:
<1> demarcates each region.The int type matrix of a definition height*width, by traversal and the way of recurrence, demarcates the sequence number in each region.
<2>, for each non-edge, gets two points as sample point, for the calculating to illumination.The current grayvalue of sample point and background gray levels are existed in structure.
<3> for the calculating of the quantity of illumination in each region, be on the whole based on
I
frame=(1-α)·I
background+α·I
shine
Wherein, I
framethe gray-scale value in this region of present frame, I
backgroundit is the gray-scale value in this region of background image.In this equation, there is α and I
shinetwo unknown quantitys.Wherein α is undated parameter, is the weighted value of the quantity of illumination, I
shineneed the quantity of illumination obtained exactly.So simultaneous two equations, form a system of equations, just can solve this two unknown quantitys.
Regenerating of <4> background.Then, to the I in each region
shineinvestigate.If it only has a peak value, namely can think that the factor affecting this region is illumination completely, so just using current scene as new background.If it has two or more peak value, so can think that the prospect (personage or vehicle) occurred in present frame has had contribution to knots modification here.So this region we just need not carry out having regenerated of background.Namely the regeneration of background is carried out by following formula:
Like this, by the piecemeal process at edge, just background can be upgraded rapidly when there being vehicle movement.
4) feature extraction and vehicle tracking
By current frame image and background image difference, just can obtain the binary image of present frame prospect, by twice expansion and twice corrosion, substantially can filter out the interference of people and bicycle.
Obtain the two-value foreground features of vehicle, the track algorithm of these classics of Meanshift just can be used to follow the tracks of vehicle, by the trajectory of vehicle stored in structure.
5) judge in violation of rules and regulations to preserve with evidence
By above-mentioned image vehicle tracking process, obtain the historical track of vehicle.Now the track of the vehicle at crossing is recalled, whether crossingly with three lines previously demarcated in the moment that red light lights detect track, if vehicle is crossing with the line of two wherein, so just can assert that this car is red light running.
Owing to having obtained the movement locus of vehicle in each frame picture before this, so we just can be easy to obtain which frame vehicle before present frame has had cross-lane to occur, and video camera all should be in the state of recorded video in whole process.We only need from the video recorded, to choose particular frame picture, as running red light for vehicle three evidence pictures in violation of rules and regulations.
Claims (4)
1., based on the detection method of making a dash across the red light of complicated high dynamic environment modeling, comprise the following steps:
1) the crossing picture that video camera photographs is demarcated;
2) gauss hybrid models is utilized to carry out background modeling to the crossing picture photographed;
3) illumination examination and analysb is carried out to the background of modeling, and carry out the renewal of background, obtain the accurate background image at current crossing;
4) trace information of crossing vehicle is obtained by background difference and track algorithm process;
5) judge whether vehicle makes a dash across the red light according to the trace information of vehicle and uncalibrated image;
Wherein, step 1) be specially: region, crossing, traffic lights region and three lines for the preservation of violation evidence are calibrated to the crossing picture that video camera photographs;
Gauss hybrid models carries out background modeling according to hsv color model, context update speed can adjust according to the similarity of current background model and previous background model, if current background model is close to previous model, the renewal speed of the background that then slows down, otherwise the renewal speed accelerating background;
In step 3) in illumination examination and analysb is carried out to the background of modeling, comprise: 31) edge of background image is extracted, edge is expanded, 32) non-edge of background image is divided into multiple subregion according to position, 33) two sample points are got in each subregion, and calculate the quantity of illumination in each subregion, 34) judge whether to carry out subregional context update according to the result of calculation of the quantity of illumination;
Gauss hybrid models can obtain accurate background image by the method detecting and eliminate a large amount of disturbed motion object;
The detection method of disturbed motion object is specially, the pixel value difference of present frame and background image same point and motion threshold limit value are compared, according to comparative result determination disturbed motion object, then the pixel being defined as disturbed motion object is carried out all disturbed motion targets that piecemeal process obtains in background image;
Step 5) be specially: whether the track detecting vehicle is crossing with three lines previously demarcated in the moment that red light lights, if described vehicle is crossing with the line of two wherein, then assert that described running red light for vehicle is in violation of rules and regulations.
2. detection method of making a dash across the red light according to claim 1, is characterized in that, the background image being defined as disturbed motion target does not carry out context update.
3. detection method of making a dash across the red light according to claim 1, it is characterized in that, step 4) be specially: binary image current frame image and background image difference being obtained present frame prospect, filtered out the interference of people and bicycle by twice expansion and twice corrosion, then use Meanshift track algorithm vehicle to be followed the tracks of to the track of vehicle obtaining current frame image to prospect.
4. based on the red light overriding detection system of complicated high dynamic environment modeling, comprise camera head, control device, camera head is connected with control device and sends to it crossing picture photographed, it is characterized in that, described control device is used for carrying out background modeling by gauss hybrid models to the crossing picture photographed, illumination examination and analysb is carried out to the background of modeling, and carry out the renewal of background, obtain the accurate background image at current crossing, the trace information of crossing vehicle is obtained by background difference and track algorithm process, judge whether vehicle makes a dash across the red light according to the trace information of vehicle and uncalibrated image,
Wherein, the crossing picture of described camera head also for photographing video camera calibrates region, crossing, traffic lights region and three lines for the preservation of violation evidence;
Described control device also carries out background modeling for gauss hybrid models according to hsv color model, context update speed can adjust according to the similarity of current background model and previous background model, if current background model is close to previous model, the renewal speed of the background that then slows down, otherwise the renewal speed accelerating background;
Described control device also for carrying out illumination examination and analysb to the background of modeling, comprising: 1) extract the edge of background image, being expanded at edge; 2) non-edge of background image is divided into multiple subregion according to position; 3) get two sample points in each subregion, and calculate the quantity of illumination in each subregion; 4) judge whether to carry out subregional context update according to the result of calculation of the quantity of illumination;
Described control device also can obtain accurate background image by the method detecting and eliminate a large amount of disturbed motion object for gauss hybrid models;
Described control device is also for comparing the pixel value difference of present frame and background image same point and motion threshold limit value, according to comparative result determination disturbed motion object, then the pixel being defined as disturbed motion object is carried out all disturbed motion targets that piecemeal process obtains in background image;
Whether the track of described control device also for detecting vehicle be crossing with three lines previously demarcated in the moment that red light lights, if described vehicle is crossing with the line of two wherein, then assert that described running red light for vehicle in violation of rules and regulations.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201310117993.6A CN103218916B (en) | 2013-04-07 | 2013-04-07 | Method and system for detecting red light running based on complex high-dynamic environmental modeling |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201310117993.6A CN103218916B (en) | 2013-04-07 | 2013-04-07 | Method and system for detecting red light running based on complex high-dynamic environmental modeling |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN103218916A CN103218916A (en) | 2013-07-24 |
| CN103218916B true CN103218916B (en) | 2015-01-21 |
Family
ID=48816656
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201310117993.6A Active CN103218916B (en) | 2013-04-07 | 2013-04-07 | Method and system for detecting red light running based on complex high-dynamic environmental modeling |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN103218916B (en) |
Families Citing this family (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2017028010A1 (en) * | 2015-08-14 | 2017-02-23 | 富士通株式会社 | Background model extracting method and apparatus and image processing device |
| CN110111577B (en) * | 2019-05-15 | 2020-11-27 | 武汉纵横智慧城市股份有限公司 | Non-motor vehicle identification method, device, equipment and storage medium based on big data |
| CN113902986A (en) * | 2021-09-14 | 2022-01-07 | 中山大学 | Method and system for fine-grained recognition of video road snow state based on deep learning |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101123039A (en) * | 2007-09-14 | 2008-02-13 | 清华大学 | Red light detection system and method based on digital camera |
| CN102496281A (en) * | 2011-12-16 | 2012-06-13 | 湖南工业大学 | Vehicle red-light violation detection method based on combination of tracking and virtual loop |
| CN102881161A (en) * | 2012-09-28 | 2013-01-16 | 武汉烽火众智数字技术有限责任公司 | Method and device for detecting moving vehicles on basis of multi-frame differences and cast shadow removal |
Family Cites Families (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20050187701A1 (en) * | 2004-02-23 | 2005-08-25 | Baney Douglas M. | Traffic communication system |
| CN101686338B (en) * | 2008-09-26 | 2013-12-25 | 索尼株式会社 | System and method for partitioning foreground and background in video |
-
2013
- 2013-04-07 CN CN201310117993.6A patent/CN103218916B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101123039A (en) * | 2007-09-14 | 2008-02-13 | 清华大学 | Red light detection system and method based on digital camera |
| CN102496281A (en) * | 2011-12-16 | 2012-06-13 | 湖南工业大学 | Vehicle red-light violation detection method based on combination of tracking and virtual loop |
| CN102881161A (en) * | 2012-09-28 | 2013-01-16 | 武汉烽火众智数字技术有限责任公司 | Method and device for detecting moving vehicles on basis of multi-frame differences and cast shadow removal |
Non-Patent Citations (1)
| Title |
|---|
| 基于视频技术的车辆违章行为检测;骆迪;《中国优秀硕士学位论文全文数据库 信息科技辑》;20100215(第2期);第I140-230页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN103218916A (en) | 2013-07-24 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109816024B (en) | A real-time vehicle logo detection method based on multi-scale feature fusion and DCNN | |
| CN103778786B (en) | A kind of break in traffic rules and regulations detection method based on remarkable vehicle part model | |
| CN105678811B (en) | A kind of human body anomaly detection method based on motion detection | |
| CN101996410B (en) | Method and system of detecting moving object under dynamic background | |
| CN104537841B (en) | Unlicensed vehicle violation detection method and detection system thereof | |
| CN103530893B (en) | Based on the foreground detection method of background subtraction and movable information under camera shake scene | |
| CN109147368A (en) | Intelligent driving control method device and electronic equipment based on lane line | |
| CN106778540B (en) | Parking detection is accurately based on the parking event detecting method of background double layer | |
| CN112241969A (en) | Target detection and tracking method, device and storage medium based on traffic surveillance video | |
| CN109190488A (en) | Front truck car door opening detection method and device based on deep learning YOLOv3 algorithm | |
| CN102592454A (en) | Intersection vehicle movement parameter measuring method based on detection of vehicle side face and road intersection line | |
| KR20210074961A (en) | Road traffic information analyzing method by using drone shooting video and device using thereof | |
| JP6700373B2 (en) | Apparatus and method for learning object image packaging for artificial intelligence of video animation | |
| CN110889328A (en) | Method, device, electronic equipment and storage medium for detecting road traffic condition | |
| CN103488993A (en) | Crowd abnormal behavior identification method based on FAST | |
| CN108550258A (en) | Vehicle queue length detection method, device, storage medium and electronic equipment | |
| CN114782915B (en) | Intelligent automobile end-to-end lane line detection system and equipment based on auxiliary supervision and knowledge distillation | |
| CN106778484A (en) | Moving vehicle tracking under traffic scene | |
| JP6681965B2 (en) | Apparatus and method for extracting learning target image for autonomous driving | |
| CN113160299B (en) | Vehicle video speed measurement method based on Kalman filtering and computer readable storage medium | |
| CN115223106A (en) | Sprinkler detection method fusing differential video sequence and convolutional neural network | |
| CN103218916B (en) | Method and system for detecting red light running based on complex high-dynamic environmental modeling | |
| CN104318760B (en) | A method and system for intelligent detection of intersection violations based on object-likeness model | |
| CN119600510B (en) | Video data driven tunnel fire detection and simulation method and system | |
| CN115909219A (en) | Scene change detection method and system based on video analysis |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant | ||
| TR01 | Transfer of patent right |
Effective date of registration: 20171010 Address after: 610000 No. 888, south section of Tianfu Road, Tianfu New District, Sichuan, Chengdu Patentee after: BUFFALO ROBOT TECHNOLOGY (CHENGDU) Co.,Ltd. Address before: 610000 Chengdu province high tech Zone (West) West source Avenue, No. 2006 Co-patentee before: Cheng Hong Patentee before: Chengdu electronics great assets management Co.,Ltd. Effective date of registration: 20171010 Address after: 610000 Chengdu province high tech Zone (West) West source Avenue, No. 2006 Co-patentee after: Cheng Hong Patentee after: Chengdu electronics great assets management Co.,Ltd. Address before: 215311 Kunshan, Jiangsu, Pakistan Town, Xueyuan Road, No. 88, No. Co-patentee before: University of Electronic Science and Technology of China Patentee before: Buffalo Robot Technology (Suzhou) Co.,Ltd. |
|
| TR01 | Transfer of patent right |